jbm@uncle.UUCP (John B. Milton) (03/14/89)
In article <1429@mtunb.ATT.COM> jcm@mtunb.UUCP (John McMillan) writes: >In article <296@carroll1.UUCP> dnewton@carroll1.UUCP (Dave Newton) writes: >>The 68010 has a built-in MMU (Memory Management Unit) which is vital to >>the UNIXpc's multi-tasking operation. >Please so inform Motorola: They've yet to realize this! ya ba dee yub >Their documentation is stupid enough to suggest they SUPPORT virtual > memory (page faults, etc.) with improved restarts in the 68010. > (I'm not sure one couldn't do the same with a 68000 -- given > the ungodsly effort I've seen for Z8000 instruction-restart -- > but it wouldn't be pretty!-) They did a very good job, considering. >I'd recommend using the OFF-chip MMU that the 3B1 comes with, and which, > alas, limits the 3B1 to 4MB VM. And which EXISTS, if that's of concern. Yes, for THIS machine I would. Although having to copy the portion of the mapping memory for a given process in and out with every context switch is time consuming, it was more than worth the alternative when this machine was designed. Now that big, high speed static RAMs are becoming available, I could see this method taken even further. >> Switching it for a 68000 would indeed >>be very bad. > ^^^ Not really fatal, if you have TWO 68000s. This was actually suggest by Motorola in an app. note, and IMPLEMENTED in the SUN-2 (or was it the SUN-1?). What you do is, when a page fault occures, hold off DTACK to the CPU that just did this nasty thing, then run on the SECOND 68000 to do the paging. When everything is okydoky, release DTACK and let the first 68000 continue. The first one doesn't know that there was a many ms delay between two CYCLES of one instruction. Of course, you have to be able to disable any watch-dog BERR timers while all this is going on. >"Fatal", yes. Bad? Thats a moral or ethical decision. (Unless you're a > politician where such things are NOPs.) Hmm, ok. >Everyone is CAPABLE of submitting answers or statements: > 1) Recognize your limits and > 2) try to exercise more courtesy -- like reading to the end of the > news listings to see if others have posted replies. The only way people LEARN is by asking questions. At first stupid maybe, questions anyway. Many wonderful things have come from some "uninformed" person asking what appears to be a stupid question. **** D O N ' T F L A M E , I N F O R M **** I think I just found a new .signature John -- John Bly Milton IV, jbm@uncle.UUCP, n8emr!uncle!jbm@osu-cis.cis.ohio-state.edu (614) h:294-4823, w:764-2933; AMPR: 44.70.0.52; Don't FLAME, inform!
guy@auspex.UUCP (Guy Harris) (03/16/89)
>Not really fatal, if you have TWO 68000s. This was actually suggest by Motorola >in an app. note, and IMPLEMENTED in the SUN-2 (or was it the SUN-1?). Neither. The original Sun-1s (not all caps, please; when used as the name of the company or in product names, it's not an acronym) had a 68000, as I remember, and didn't run an OS that supported VM. A Sun-1 could be upgraded to an '010 machine; the Sun-2's were all '010 machines. The '010 machines didn't need that trick. As I remember, Apollo and Masscomp *did* use the "two 68000s" trick.